Varying geom_hline for each facet_wrap plot

Question

For each of the nine plots, I am trying to create two horizontal lines of the same color as "Method" where the y-intercept is the value of "Probability of Go" (ignore that these are not valid probabilities) when the "true landmark" is 0.65. For example, the top left plot would have a blue horizontal line at about y=6.25 and a red horizontal line at about y=5.2. Is there a simple way to do this without manually specifying the y-intercept for each of the 9 plots and for each of the 2 methods? My code right now is as follows:

true_landmark=seq(0.15,0.85,by=0.05);pt=6;theta=0.65
L=length(true_landmark)
type <- rep(c(rep("Exponential",L*2),rep("Log-Normal",L*2),rep("Weibull",L*2)),3)
method <- rep(c(rep("Landmark PFS",L),rep("RMST",L)),3*3)
n <- c(rep(20,L*2*3),rep(30,L*2*3),rep(40,L*2*3))
value <- replicate(18,cumsum(runif(15)))
landmark <- rep(true_landmark,2*3*3)  
df <- data.frame(type,method,n,value,landmark)
df$n <- factor(df$n,labels=c(expression(n==20),expression(n==30),expression(n==40)))
df$type <- factor(df$type,
                  labels=c("Exponential",
                           expression(paste("Log-Normal (",sigma==1,")")),
                           expression(paste("Weibull (",k==0.5,")"))))

ggplot(df, aes(landmark,value,color=method)) +
  geom_line(alpha=0.5)+
  geom_point(shape=19,alpha=0.5)+
  geom_blank()+
  scale_x_continuous(name =paste("True Landmark PFS at", pt, "Months"), breaks=seq(true_landmark[1],true_landmark[length(true_landmark)], 0.1))+
  ylab(label="Probability of Go")+
  geom_vline(xintercept = theta, color="black", linetype="dashed")+
  #geom_text(aes(x=theta-0.02, label=paste("Target value =", theta), y=0.30), angle=90, size=4, color="black")+
  facet_grid(n~type,labeller=label_parsed)+
  guides(color=guide_legend(title="Method"))+
  theme(plot.caption=element_text(hjust=0))+
  labs(caption=paste("Go: Posterior prob (True PFS/RMST at", pt, "month > target|data)", ">", P_T,"\nDashed line indicates target landmark PFS/RMST value"))

Answer 1

You have made things a bit more difficult for yourself by leaving value as an array outside of the data frame (notice that although you include it when making df, as an array it just creates a bunch of columns called X1, X2, etc). You can solve the problem like this:

ggplot(df, aes(landmark, value, color = method)) +
  geom_line(alpha = 0.5)+
  geom_point(shape = 19, alpha = 0.5) +
  geom_blank() +
  geom_hline(data = df[df$landmark == 0.65,],
             aes(yintercept = value[df$landmark == 0.65], color = method)) +
  scale_x_continuous(name = paste("True Landmark PFS at", pt, "Months"), 
                     breaks = seq(true_landmark[1], 
                                  true_landmark[length(true_landmark)], 0.1)) +
  ylab(label="Probability of Go") +
  geom_vline(xintercept = theta, color = "black", linetype = "dashed") +
  facet_grid(n~type,labeller = label_parsed)+
  guides(color = guide_legend(title = "Method")) +
  theme(plot.caption = element_text(hjust = 0)) +
  labs(caption = paste("Go: Posterior prob (True PFS/RMST at", pt, 
                      "month > target|data)", ">",
                      "\nDashed line indicates target landmark PFS/RMST value"))